The Tradition Machine Learning Approach with various Pros and Cons

The Tradition Machine Learning Approach with pros and cons and applications

1. Module 1 Introducing Deep Learning The Traditional Machine Learning Approach

2. The Traditional Machine Learning Approach In Machine Learning 1. It is an automatic Process

3. The Traditional Machine Learning Approach 2. These Models are nothing but programs generated by algorithms without being explicitly programmed by any human being. 3. These models continuously evolve into better models automatically as the amount of quality data increases.

4. The Traditional Machine Learning Approach The traditional ML approach refers to the way machine learning problems are solved before deep learning became popular . It usually involves : Data Collection – Gathering structured data (like CSV, Excel, SQL databases). Feature Engineering – Selecting or designing meaningful input variables (features) that represent the data. Model Selection – Choosing an ML algorithm (e.g., Decision Tree, SVM, Logistic Regression). Training – Fitting the model on the dataset. Evaluation – Testing accuracy, precision, recall, etc. Prediction/Deployment – Using the trained model on new unseen data .

5. The Traditional Machine Learning Approach Example 1: Email Spam Detection Data: Emails with text and labels (spam or not spam). Feature Engineering: Extract features like word frequency (e.g., "free", "win", "money"), length of email, number of links. Model Selection: Naïve Bayes or Logistic Regression. Training: Train model on 80% of the dataset. Evaluation: Test model on 20% to see accuracy. Prediction: New email → Model predicts "Spam" or "Not Spam".

6. The Traditional Machine Learning Approach Example 2: Predicting House Prices Data : Dataset with columns like number of rooms, location, square footage, and price. Feature Engineering: Maybe create new features like "price per square foot". Model Selection: Linear Regression or Random Forest. Training: Train model using historical house data. Evaluation: Measure prediction error (Mean Squared Error). Prediction: Input → 3BHK, 1500 sqft in City A → Output → Predicted price.

7. The Traditional Machine Learning Approach Example 3: Medical Diagnosis Data: Patient records with blood pressure, cholesterol, sugar level, and whether they have diabetes. Feature Engineering: Normalize values, add BMI, combine features. Model Selection: Support Vector Machine (SVM) or Decision Tree. Training & Evaluation: Train on patient history data. Prediction: For a new patient, model predicts "Diabetic" or "Non-Diabetic".

8. The Traditional Machine Learning Approach Compared with Deep Learning Traditional ML: Relies heavily on feature engineering (manual design of features ). Deep Learning: Automatically learns features from raw data (e.g., images, text).

9. THANK YOU